Given $m$ vectors, is the sum of all possible inner products between them nonnegative?

51 Views Asked by At

I'm trying to solve the following problem from Linear Algebra Done Right, by Sheldon Axler.

Prove or give a counterexample: if $v_1, \dots , v_m \in V$, then $$\sum_{j = 1}^{m}\sum_{k = 1}^{m}\langle v_j, v_k \rangle \geq 0.$$

I don't know if the statement is true, so all I've done so far is try to manipulate the expression: $$\sum_{j = 1}^{m}\sum_{k = 1}^{m}\langle v_j, v_k \rangle = \sum_{j = k}^{}\sum_{k = 1}^{m} \langle v_j, v_k \rangle + \sum_{j \neq k}^{}\sum_{k = 1}^{m}\langle v_j, v_k \rangle = \sum_{j = 1}^{m} \langle v_j, v_j \rangle + 2 \sum_{k > j} \Re(\langle v_j, v_k \rangle),$$ whose first term is nonnegative, by defintion of inner product. However, I can't say anything about the second term. Any help would be appreciated.

1

There are 1 best solutions below

0
On BEST ANSWER

The trick here comes down to the linearity of the dot-product: $$ \sum_{j = 1}^{m}\sum_{k = 1}^{m}\left\langle v_j, v_k \right\rangle = \sum_{j = 1}^{m}\left\langle v_j, \sum_{k = 1}^{m}v_k \right\rangle = \left\langle\sum_{j = 1}^{m} v_j, \sum_{k = 1}^{m}v_k \right\rangle $$ That is, the expression can be written as $\langle w,w\rangle$ where $w = \sum_{j = 1}^{m} v_j$.